Forecasting Daily Chlorophyll a Concentration during the Spring Phytoplankton Bloom Period in Xiangxi Bay of the Three-Gorges Reservoir by Means of a Recurrent Artificial Neural Network

A recurrent artificial neural network was used for 0-and 7-days-ahead forecasting of daily spring phytoplankton bloom dynamics in Xiangxi Bay of Three-Gorges Reservoir with meteorological, hydrological, and limnological parameters as input variables. Daily data from the depth of 0.5 m was used to train the model, and data from the depth of 2.0 m was used to validate the calibrated model. The trained model achieved reasonable accuracy in predicting the daily dynamics of chlorophyll a both in 0-and 7-days-ahead forecasting. In 0-day-ahead forecasting, the R-2 values of observed and predicted data were 0.85 for training and 0.89 for validating. In 7-days-ahead forecasting, the R-2 values of training and validating were 0.68 and 0.66, respectively. Sensitivity analysis indicated that most ecological relationships between chlorophyll a and input environmental variables in 0-and 7-days-ahead models were reasonable. In the 0-day model, Secchi depth, water temperature, and dissolved silicate were the most important factors influencing the daily dynamics of chlorophyll a. And in 7-days-ahead predicting model, chlorophyll a was sensitive to most environmental variables except water level, DO, and NH3N.

SOA-based actively mode-locked fiber ring laser by forward injecting an external pulse train

An actively mode-locked fiber ring laser based on cross-gain modulation (XGM) in a semiconductor optical amplifier (SOA) is demonstrated to operate stably with a simple configuration. By forward injecting an easily-generated external pulse train, the mode-locked fiber laser can generate an optical-pulse sequence with pulsewidth about 6 ps and average output power about 7.9 mW. The output pulses show an ultra-low RMS jitter about 70.7 fs measured by a RF spectrum analyzer. The use of the proposed forward-injection configuration can realize the repetition-rate tunability from I to 15 GHz for the generated optical-pulse sequences. By employing a wavelength-tunable optical band-pass filter in the laser cavity, the operation wavelength of the designed SOA-based actively mode-locked fiber laser can be tuned continuously in a wide span between 1528 and 1565 nm. The parameters of external-injection optical pulses are studied experimentally to optimize the mode-locked fiber laser. (C) 2009 Elsevier B.V. All rights reserved.

Determination of interface properties between micron-thick metal film and ceramic substrate using peel test

Peel test measurements have been performed to estimate both the interface toughness and the separation strength between copper thin film and Al2O3 substrate with film thicknesses ranging between 1 and 15 mu m. An inverse analysis based on the artificial neural network method is adopted to determine the interface parameters. The interface parameters are characterized by the cohesive zone (CZ) model. The results of finite element simulations based on the strain gradient plasticity theory are used to train the artificial neural network. Using both the trained neural network and the experimental measurements for one test result, both the interface toughness and the separation strength are determined. Finally, the finite element predictions adopting the determined interface parameters are performed for the other film thickness cases, and are in agreement with the experimental results.

基于神经网络的激光熔覆高度预测

激光成形过程中,对熔覆高度进行实时检测,从而实现熔覆高度闭环控制是成形高质量零件的保证。激光成形过程是一个多参数耦合的非线性过程,大量激光参数对成形熔覆表面质量具有重要影响。在分析激光参数对熔覆高度影响的基础上,建立利用激光工艺参数预测熔覆高度的误差反向传播(Backpropagation,BP)神经网络模型,完成了网络算法设计。通过激光成形试验采集样本,利用训练样本对所建立的网络进行训练,完成网络输入输出高度映射关系,并利用测试样本对所训练的网络进行检验。仿真试验表明,神经网络熔覆高度预测模型具有很高的精度,验证了该预测模型在理论和实践上的可行性与有效性。神经网络熔覆高度预测模型为实现激光加工过程熔覆高度实时预测与闭环控制打下基础,对提高成形产品质量具有重要意义。

The Relationship between Decision Trees and the Scale of Train Data Set

Decision Trees need train samples in the train data set to get classification rules. If the number of train data was too small, the important information might be missed and thus the model could not explain the classification rules of data. While it is not affirmative that large scale of train data set can get well model. This Paper analysis the relationship between decision trees and the train data scale. We use nine decision tree algorithms to experiment the accuracy, complexity and robustness of decision tree algorithms. Some results are demonstrated.

Spectral effects for an ultrashort pulse train propagating in a two-level atom medium

We investigate the spectra of a femtosecond pulse train propagating in a resonant two-level atom (TLA) medium. it is found that higher spectral components can be produced even for a 2 pi femtosecond pulse train. Furthermore, the spectral effects depend crucially on both the relative shift phi and the delay time tau between the successive pulses of the femtosecond pulse train.

Coherent population accumulations of multiphoton transitions induced by an ultrashort pulse train in polar molecules

Coherent population accumulations of multiphoton transitions induced by an ultrashort pulse train in a two-level polar molecule are investigated theoretically by solving the density-matrix equations without invoking any of the standard approximations. It is shown due to the effects of permanent dipole moments, that the population accumulation of multiphoton transitions can be obtained in the polar molecule. Moreover, the population accumulations depend crucially on the relative phase between two sequential pulses, and the period in which the maximum population accumulation occurs is 2 pi/N in N-photon transitions.

The ontogeny of echolocation in a Yangtze finless porpoise (Neophocaena phocaenoides asiaeorientalis)

Acoustic and concurrent behavioral data from one neonatal male Yangtze finless porpoise (Neophocaena phocaenoides asiaeorientalis) in captivity were presented. The calf click train was first recorded at 22 days postnatal, and the frequency of hydrophone-exploration behavior with head scanning motions in conjunction with emissions of click trains by the calf increased gradually with age. The echolocation clicks in the first recorded click train were indistinguishable from those of adults. Calf echolocation trains were found to decrease in maximum click-repetition rate, duration, and number of clicks per train with age while the. minimum click-repetition rate remained more consistent. (c) 2007 Acoustical Society of America.

Optical spectral relation between the master and the frequency-locked slave lasers

In this paper, the spectral relation between the master and the frequency-locked slave laser (FLSL) is investigated by the conventional technique of optical intensity modulation and optical heterodyne. Experimentally, we demonstrate that under complete and stable locking condition, the lightwave of the FLSL and the sidebands of the master laser produced by the optical intensity modulation are perfectly coherent (frequency coherence). Referring to our recent studies, the lightwave of the master laser and its corresponding sidebands are also perfectly coherent. Additionally, the spectral structures of two perfectly coherent lightwaves are identical in the level of wave train. Therefore, we indirectly verify that the spectral structures of the FLSL and the master laser are identical in the level of wave train.

Formal specification and refinement of a safe train control function

Motivated by the design and development challenges of the BART case study, an approach for developing and analyzing a formal model for reactive systems is presented. The approach makes use of a domain specific language for specifying control algorithms able to satisfy competing properties such as safety and optimality. The domain language, called SPC, offers several key abstractions such as the state, the profile, and the constraint to facilitate problem specification. Using a high-level program transformation system such as HATS being developed at the University of Nebraska at Omaha, specifications in this modelling language can be transformed to ML code. The resulting executable specification can be further refined by applying generic transformations to the abstractions provided by the domain language. Problem dependent transformations utilizing the domain specific knowledge and properties may also be applied. The result is a significantly more efficient implementation which can be used for simulation and gaining deeper insight into design decisions and various control policies. The correctness of transformations can be established using a rewrite-rule based induction theorem prover Rewrite Rule Laboratory developed at the University of New Mexico.

Periodic bubble emission and appearance of an ordered bubble sequence (train) during condensation in a single microchannel

Condensation of steam in a single microchannel, silicon test section was investigated visually at low flow rates. The microchannel was rectangular in cross-section with a depth of 30 pm, a width of 800 mu m and a length of 5.0 mm, covered with a Pyrex glass to allow for visualization of the bubble formation process. By varying the cooling rate during condensation of the saturated water vapor, it was possible to control the shape, size and frequency of the bubbles formed. At low cooling rates using only natural air convection from the ambient environment, the flow pattern in the microchannel consisted of a nearly stable elongated bubble attached upstream (near the inlet) that pinched off into a train of elliptical bubbles downstream of the elongated bubble. It was observed that these elliptical bubbles were emitted periodically from the tip of the elongated bubble at a high frequency, with smaller size than the channel width. The shape of the emitted bubbles underwent modifications shortly after their generation until finally becoming a stable vertical ellipse, maintaining its shape and size as it flowed downstream at a constant speed. These periodically emitted elliptical bubbles thus formed an ordered bubble sequence (train). At higher cooling rates using chilled water in a copper heat sink attached to the test section, the bubble formation frequency increased significantly while the bubble size decreased, all the while forming a perfect bubble train flowing downstream of the microchannel. The emitted bubbles in this case immediately formed into a circular shape without any further modification after their separation from the elongated bubble upstream. The present study suggests that a method for controlling the size and generation frequency of microbubbles could be so developed, which may be of interest for microfluidic applications. The breakup of the elongated bubble is caused by the large Weber number at the tip of the elongated bubble induced by the maximum vapor velocity at the centerline of the microchannel inside the elongated bubble and the smaller surface tension force of water at the tip of the elongated bubble.

Generation of 10 GHz, 1.9 ps optical pulse train using semiconductor optical amplifier and silica-based highly nonlinear fiber

We propose a simple approach to generate a high quality 10 GHz 1.9 ps optical pulse train using a semiconductor optical amplifier and silica-based highly nonlinear fiber. An optical pulse generator based on our proposed scheme is easy to set up with commercially available optical components. A 10 GHz, 1.9 ps optical pulse train is obtained with timing jitter as low as 60 fs over the frequency range 10 Hz-1 MHz. With a wavelength tunable CW laser, a wide wavelength tunable span can be achieved over the entire C band. The proposed optical pulse generator also can operate at different repetition rates from 3 to 10 GHz.